Article pubs.acs.org/Organometallics
Radical Mechanisms in the Reaction of Organic Halides with Diiminepyridine Cobalt Complexes Di Zhu,† Ilia Korobkov,‡ and Peter H. M. Budzelaar*,† †
Department of Chemistry, University of Manitoba, Winnipeg, MB R3T 2N2, Canada Department of Chemistry, University of Ottawa, Ottawa, ON K1N 6N5, Canada
‡
S Supporting Information *
ABSTRACT: The formally Co(0) complex LCo(N2) (L = 2,6-bis(2,6-dimethylphenyliminoethyl)pyridine) can be prepared via either Na/Hg reduction of LCoCl2 or hydrogenolysis of LCoCH2SiMe3. In the latter reaction, LCoH could be trapped by reaction with NCC6H4-4-Cl to give LCoN CHC6H4-4-Cl. LCo(N2) reacts with many alkyl and aryl halides RX, including aryl chlorides, to give a mixture of LCoR and LCoX in a halogen atom abstraction mechanism. Intermediacy of free alkyl and aryl radicals is confirmed by the ring-opening of cyclopropylmethyl to crotyl, and the rearrangement of 2,4,6-tBu3C6H2 to 3,5-tBu2C6H3CMe2CH2, before binding to Co. The organocobalt species generated in this way react further with activated halides R′X (alkyl iodides; allyl and benzyl halides) to give cross-coupling products RR′ in what is most likely again a halogen abstraction mechanism. DFT studies support the proposed radical pathways for both steps. MeI couples smoothly with LCoCH2SiMe3 to give LCoI and CH3CH2SiMe3, but the analogous reaction of tBuI leads in part to radical attack at the 3 and 4 positions of the pyridine ring to form (tBu2-L)CoI and (tBu2-L)CoI2.
■
INTRODUCTION In recent years, redox-active ligands have been intensively studied due to the unusual properties they confer on their metal complexes. Popular classes of such ligands are α-diimines,1 α-imino-ketones,2 and diiminepyridines (DIP);3 the latter are especially attractive because they bind strongly to transition metals and provide considerable steric protection. One intriguing aspect of the chemistry of redox-active ligands is the possibility of “ennobling” light transition metals,4 making them behave more like their heavier congeners. For example, DIP complexes of iron catalyze hydrogenation and hydrosilylation,5 in a manner more typical for the platinum metals. Whereas second- and third-row transition metals mostly exhibit low-spin states and 2e redox reactions, first-row transition metals tend to display high-spin states and 1e redox steps. However, the possibility of having transition-metal-centered unpaired electrons antiferromagnetically (AF) coupled to ligandcentered electrons opens up the possibility of having 2e steps where both ligand and metal are oxidized or reduced in a “coupled” fashion. On the other hand, the flexible electronic structure of complexes of redox-active ligands might also allow for new reaction modes with significant potential in synthesis and catalysis. One of the most important redox reactions in organic synthesis is the breaking of carbon−halogen bonds. This is a key step in the catalyzed formation of C−C, C−N, and C−S bonds; in addition, C−X cleavage is required in the disposal of CFCs and similarly harmful environmental contaminants. The most common mechanisms of C−X cleavage are6 © 2012 American Chemical Society
(a) via SN2-like nucleophilic attack by an electron-rich metal center (mostly for alkyl halides)
(b) via concerted addition involving a three-center transition state (mostly for aryl halides)
(c) via radical mechanisms (usually for activated alkyl halides) In each of these, one typically obtains a product that has the halide and the carbon fragment attached to the same metal atom (“mononuclear oxidative addition”) in an overall 2e oxidation step (Scheme 1, A). Cases where the halide and the organic fragment end up on two different metal atoms in two separate 1e oxidation steps (“binuclear oxidative addition”: Scheme 1, B) are much more rare. Most of them involve alkyl halides,7−10 presumably because the Csp3−X bond is weaker than the Csp2−X bond as Received: March 5, 2012 Published: May 15, 2012 3958
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
Article
(R = CH2SiMe3) in THF or benzene with a mixture of H2 and N2.11 Formation of (1)Co(N2) from (1)CoCl2 and Na/Hg likely involves straightforward reduction andat some stage capture of a dinitrogen molecule. In contrast, the mechanism by which (1)Co(N2) forms from (1)CoR and H2/N2 seems less obvious. (3)CoR reacts quickly with H2 to give (3)CoH, which is fairly stable at room temperature and is efficient at catalyzing olefin hydrogenation,17 a reaction that likely involves hydrogenolysis of (3)Co(alkyl) intermediates. Thus, one would expect treatment of (1)CoR with H2 to initially produce (1)CoH. This hydride appears to be much less stable than (3)CoH,17b and we have never directly observed it in 1H NMR spectra. However, a hydride intermediate could be trapped (Scheme 3) by treating (1)CoR with H2 in the presence of
Scheme 1. Schematic Representation of (A) Mononuclear and (B) Binuclear Oxidative Additions
well as for steric reasons. In fact, prior to our own recent communication,11 the only example of binuclear oxidative addition of an aryl halide was reported about 45 years ago,9a when Halpern and co-workers observed the reaction of the activated halide 2-iodopyridine with Co(CN)53− to give a mixture of ICo(CN)53− and 2-C5H4NCo(CN)53−. Redox-active ligands can make electrons available during metal-centered reactivity. This has been exploited, for example, by Soper in his study of Co−C and C−C bond formation at CoIII centers bearing amidophenolate ligands.12 The DIP ligand has two low-lying π-acceptor orbitals and can accept up to three electrons in its extended π-system.13 Thus, it should similarly be able to function as an electron reservoir during reactions at the metal center. Indeed, the group of Chirik observed oxidative addition of C−X and even C−O bonds to (DIP)Fe(0) complexes in what seems to be a traditional mononuclear oxidative addition reaction, although in many products the carbon-containing fragment had been “lost” from the metal.7b We decided to explore in some detail the potential of the related (DIP)Co(0) and (DIP)CoI complexes in oxidative addition reactions. For (DIP)Co(0), we found that surprisingly binuclear oxidative addition predominates.11 Interestingly, the reaction works best for aryl chlorides, which are normally more difficult to activate than the corresponding bromides or iodides. We then reported that (DIP)CoI(aryl) complexes undergo C−C coupling with benzyl halides.14 In the present paper, we provide full details of this remarkable radical-mediated chemistry, including an exploration of the scope for C−C coupling reactions. Interestingly, the group of Chan very recently reported binuclear oxidative addition to iridium(III) porphyrin complexes,15 so it appears that this type of reaction is not restricted to first-row transition metals.
Scheme 3. Trapping of (1)CoH with PhCCPh and 4-NCC6H4Cl
PhCCPh [to give (1)CoC(Ph)CHPh] or 4-NCC6H4Cl [to give (1)CoNCHC6H4Cl]. Both species were easily identified by 1H NMR spectroscopy. The former could not be obtained pure but always contained either some unreacted (1)CoR or (1)CoCH(Ph)CH2Ph18 resulting from over-reduction. However, ketimide complex (1)CoN CHC6H4Cl could be isolated and characterized by single-crystal X-ray diffraction. The structure (Figure 1) shows the square-planar Co coordination environment typical of DIP CoI complexes, and a near-linear Co−NC arrangement suggesting considerable ionic character in the Co−N bond.19 The N4−C41−C42 angle of 124.3(6)° leaves little doubt that this is a ketimide rather than a nitrile complex; in addition, the HCN resonance is prominently visible in the 1H NMR spectrum at 9.62 ppm (Figure S6). The above results strongly suggest that the initial product formed from (1)CoR and H2 is (1)CoH or possibly (1)Co(H)(N2). However, the route via which this transforms into (1)Co(N2) is not clear at present. Steric hindrance appears to slow formation of the N2 complex, since conversion of (2)CoH to (2)Co(N2) takes about 30 min, while (3)CoH is stable for a day or more at room temperature. This may be taken as an indication for bimolecular elimination of H2, e.g., via the path shown in Scheme 4. Steric hindrance would obviously slow the intermolecular H transfer step. Preliminary DFT calculations on strongly simplified models support easy transfer of hydrogen from LCo(N2)(H) to LCoH, but accurate calculations of the complete disproportionation path for the real ligands 1−3 are currently not feasible. Cleavage of C−X Bonds by LCo(N2) Complexes. Many alkyl and aryl chlorides, bromides, and iodides react smoothly with green (1)Co(N2) to give a purple mixture of (1)CoR and (1)CoX. While the idealized stoichiometry for this reaction would be eq 1, we typically obtained the products in a ratio (1)CoR:(1)CoX < 1:1, as detailed below.11
■
RESULTS AND DISCUSSION The choice of DIP ligand turned out to be critical in the present work. Most reactions have been carried out with the 2, 6-Me2C6H3-substituted ligand 1 (Scheme 2). In several cases, Scheme 2. DIP Ligands Used in This Work
the analogous but more sterically hindered 2,6-Et2C6H3 (2) and 2,6-iPr2C6H3 (3) ligands were also tested. Unless otherwise stated, reactions mentioned in the text refer to ligand 1. Generation of (1)Co(N2). The starting material (1)Co(N2) for this study can be generated via reduction of (1)CoCl2 with Na/Hg as described by Chirik.16 However, we found it more convenient to generate the same species via reaction of (1)CoR
2LCo(N2) + ArX → LCoAr + LCoX + 2N2 3959
(1)
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
Article
Table 1. Influence of Ligand on C−Cl Cleavage of 4-ClC6H4CF3 by LCo(N2)a ligand
LCoAr/LCoCl
completion timeb
1 2 3
0.77 0.43 0.22c
2 > 3. For
In contrast to tBuI, AdI (entry 20) did not produce any identifiable organocobalt complex; the only recognizable product was (1)CoI. 3960
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
Article
Table 2. Reaction of (1)Co(N2) with Organic Halidesa,b entry
halide RX
(1)CoR/(1) CoXc
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
PhBr PhCl 2,4,6-tBu3C6H2Br 2,6-Me2C6H3Cl p-MeCOC6H4Cl p-CF3C6H4Cl p-ClC6H4Cl p-MeC6H4Cl 2,6-Cl2C5H3N n-C8H17F (n-C6F13)CHCH2 CH2CHCH2Cl (cy-C3H5)CH2Cl Br(CH2)4CHCH2 MeI i BuBr i PrCl t BuI t BuCl AdI
0.25 0.59 0.27e ∼0 0.91 0.77 0.59 0.59 1.0 n.r. 0.50 1.0 0.54f 0.40g 0.71h 0.48i 0.20j 0.17i,k 0.08i ∼0l
yield (1)CoR, % rxn timed n.d.m n.d.m 29 n.d.m 71 89 48 38 88
1 min hours seconds days 30 min seconds seconds hours seconds
25 88 48 35 n.d.m 44 n.d.m n.d.m 8 ∼0
minutes seconds seconds seconds seconds seconds minutes seconds minutes seconds
Unactivated C−F bonds do not react, but the allylic fluoride C6F13CHCH2 produced (1)Co(CH2CHC6F12) in fair yield. On the basis of 1H NMR parameters we believe this most likely contains a σ-bound allyl moiety (σ-CH2CHC(F)C5F11): the complex exhibits the low-field pyridine H4 resonance characteristic of square-planar LCoZ complexes, whereas the η3-allyl and η3-crotyl complexes mentioned above do not. Pure (1)Co(η3-allyl) was prepared independently from (1)CoCl2 and allylmagnesium chloride, and the structure was determined by single-crystal X-ray diffraction. Its molecular structure (Figure 2) is best described as a distorted square pyramid, in which the allyl group occupies an equatorial (C41) and an apical (C43) position. The C41−C42−C43 angle of 125.0(5)° is somewhat larger than usual for π-allyl complexes, but very similar to that reported for (3)Fe(η3-allyl).7b In view of the dynamic behavior of the complex in solution (vide inf ra), we cannot rule out the presence of minor disorder in the allyl group, and therefore the bond lengths and angles in the allyl fragment should be treated with caution. Solution NMR data for this π-allyl complex clearly reveal its fluxional character. At low temperature (−60 °C), the 1H NMR spectrum shows different “left” and “right” sides (pyridine H3 and H5 are inequivalent) but equivalent “top” and “bottom” sides (only two Me signals for the 2,6-Me2C6H3 groups). Also, the allyl group shows three resonances (center, syn, and anti protons) in the ratio 1:2:2. If we assume the most stable structure in solution corresponds to the solid-state one, this means that even at low temperature there is fast exchange between the two types of square pyramids (Scheme 5, reaction A). On increasing the temperature, left−right exchange becomes visible and the pyridine H3 and H5 signals broaden and coalesce, as do the two xylyl Me peaks. At somewhat higher temperature, also the allyl syn and anti signals begin to broaden and coalesce. However, fitting of the exchange-broadened spectra clearly shows that these two exchanges are distinct processes, with rates that differ by a factor of 10−100 over the range of temperatures where both can be fitted satisfactorily, the syn/anti exchange always being slowest. This suggests that the left−right exchange is an in-place rotation of the allyl group maintaining its π-coordination (Scheme 5, reaction B), while the syn/anti exchange involves interconversion of σ- and π-bound allyl groups (Scheme 5, reaction C). The activation parameters obtained from the fitted exchange rates are shown in Scheme 4 (for full details see the SI); not surprisingly, the resulting entropy of activation for σ/π exchange is somewhat larger than for in-place rotation, although the difference may not be statistically significant. Mechanism of C−X Cleavage. As already mentioned in our preliminary communication, we believe that the actual C−X cleavage reaction follows a radical path for all or most substrates tested. On the experimental side, this is supported by the observation of radical-rearrangement products from 2,4,6-tBu3C6H2Br and cyclopropylmethyl chloride. DFT studies also support such a mechanism. Despite extensive searches, we could not locate any transition states for conventional “side-on”
a
Reaction conditions: (1)Co(N2) generated from (1)CoCH2SiMe3 (27 μmol) and H2 (2.0 mL) in 0.4 mL of C6D6; combined with ArX (27 μmol) or RX (14 μmol). bEntries 1−11 and 15 were already reported in our communication.11 cFrom 1H NMR, pyridine H4 peaks; estimated error margin ≈ 5%. dQualitative indication. eOrganocobalt product: R = CH2CMe2-3,5-tBu2C6H3 fOrganocobalt product: R = πCH2CHCHMe. gOrganocobalt product: R = primary alkyl; see SI for discussion.26 hPy H4 resonances of (1)CoMe and (1)CoI overlap; ratio determined from NCCH3 peaks. iOrganocobalt product: R = CH 2 CHMe 2 . j Organocobalt product: R = CH2CH2CH3 (a second diamagnetic cobalt(I) complex was also detected but could not be identified with certainty). kIn addition, ligand attack products were observed: (1-tBu2)CoI:(1)CoI ≈ 0.16. lAn unidentified diamagnetic cobalt(I) complex was also detected. mConversion of RX (and hence yield of (1)CoR) could not be determined from 1 H NMR.
2,4,6- t Bu 3 C 6 H 2 Br (entry 3) formed (1)CoCH 2 CMe 2 (3,5-tBu2)C6H3.23 The nature of this product was verified by comparison with (1)CoCH2CMe2C6H5, independently prepared from (1)CoCl2 and C6H5CMe2CH2MgCl. Formation of the alkyl likely involves rearrangement of the intermediate aryl free radical.24
Cyclopropylmethyl chloride (entry 13) gave the η3-crotyl complex as the only organocobalt complex. Its structure could be assigned via comparison with the independently prepared η3-allyl complex (entry 12; vide inf ra). Opening of an intermediate cyclopropylmethyl radical25 appears to be a reasonable explanation. 3961
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
Article
Figure 2. X-ray structure of (1)Co(η3-allyl) (thermal ellipsoids drawn at 30% probability, hydrogen atoms omitted for clarity). Selected bond distances (Å) and angles (deg): Co(1)−N(1): 1.826(3); Co(1)−N(2): 1.992(3); Co(1)−N(3): 1.925(3); Co(1)−C(41): 2.070(4); Co(1)−C(42): 2.016; Co(1)−C(43): 2.131(5); C(4)−C(9): 1.414(5); N(3)−C(9): 1.338(4); N(1)−C(4): 1.378(4); N(1)−C(8): 1.368(4); C(8)−C(10): 1.425(5); N(2)−C(10): 1.320(4); C(42)−C(43): 1.346(7); C(41)−C(42): 1.381(7); N(1)−Co(1)−N(3): 80.57(12); N(1)−Co(1)−N(2): 79.14(12); N(3)−Co(1)−N(2): 152.96(12); N(1)−Co(1)−C(43): 122.20(18); N(1)−Co(1)−C(42): 153.5(2); N(1)−Co(1)−C(41): 166.76(19); C(41)−Co(1)−C(43): 70.3(2); C(43)−C(42)−C(41): 125.0(5).
Scheme 5. Fluxional Behavior of (1)Co(η3-C3H5)a
a
Activation parameters (1σ error limits) obtained from Arrhenius plots to fitted exchange rates.
attack of Co on the C−X bond of aryl halides. All searches eventually converged to saddle points where only the halide is
close to the cobalt atom, and the imaginary mode is predominantly a C−X stretch (Scheme 6). 3962
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
Article
Scheme 6. Schematic Representation of Br Abstraction TSa
both reducing electrons come from the ligand, and the electronic structure is completely changed, with the product containing high-spin FeII and an “innocent” DIP ligand.29 In (1)Co(N2), the metal is low-spin (diamagnetic) CoI, with only a small amount of spin polarization at the metal (Figure 3, left); the single unpaired electron resides in a ligand π*-orbital.28 The halide abstraction product (1)CoX has low-spin CoII AF coupled to a ligand-centered unpaired electron20 (Figure 3, right). Therefore, to proceed from reactant to product one could imagine simple transfer of a single Co d electron to the aryl halide; its remaining partner could then AF couple to the ligand-centered electron to smoothly give the product. However, this is not what we find. At the transition state for C−X bond cleavage, the metal has undergone a spin-flip to high-spin CoI (two β electrons); an amount of spin density corresponding to one α electron is spread out over both the DIP ligand and the aryl halide, apparently in the process of being transferred from one to the other (Figure 3, middle). The change in metal spin state at this point is also evident from the increase in Co−N bond lengths, by about 0.1 Å for Co−Npy and nearly 0.2 Å for Co−Nim. Finally, the reduction in DIP ligand π* population results in an increase of the Cpy−Cim bond lengths and a shortening of the CN bonds. After the electron has been transferred and the phenyl radical ejected, at some point one of the two metal-centered unpaired electrons has to flip and transfer to the ligand to produce the ground state of the product LCoX. These movements and flips of electrons are summarized in Scheme 8.
a
Bond lengths in Å, angle in deg; the double arrow represents the imaginary mode.
Consistent with the observed relative reactivity of the various halides, the free energy barrier calculated for PhBr (19.0 kcal/mol) is substantially lower than that for PhCl (23.2 kcal/mol),11 and the magnitude of both barriers is consistent with reactions happening around room temperature (1 min for PhBr, hours for PhCl). One possible mechanistic picture for the complete reaction is shown in Scheme 7. The Scheme 7. Proposed Reaction Mechanism for Reaction of LCo(N2) with Aryl Halides
aryl halide first displaces N2 from the metal (the retarding effect of THF, mentioned earlier, can be ascribed to competition between THF and ArX for the empty site at Co). Then, C−X cleavage results in expulsion of an aryl radical, which will independently make its way to a second Co center. Such radicals are highly reactive, and the low yield for some substrates may be due to the ease with which these radicals undergo side reactions. Also, the lower yields for bromides and iodides (compared to chlorides) might be due to the larger concentration of radicals formed, leading to a higher probability of radical combination.27 It is interesting to contrast this reaction mechanism with the one for addition of vinyl acetate to (3)Fe(N2)2, where Chirik explained formation of the “traditional” product (3)Fe(C2H3)(OAc) via standard mononuclear oxidative addition.7b In (3)Fe(N2)2, the metal has the actual oxidation state +II and is intermediate-spin;28 its two unpaired electrons each AF couple to a ligand-centered electron. In the oxidative addition,
Scheme 8. Movements and Flips of Electrons during Reaction of (1)Co(N2) with PhBr
Thus, the ArX halide abstraction process appears to be more complicated than one might at first imagine, involving a
Figure 3. Spin density plots for (1)Co(N2), the Ph−Br cleavage transition state, and the final products (1)CoBr and Ph·. The sign of the spin population on Co (nCo) is given relative to that of density on the ligand. Bond lengths in Å. 3963
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
Article
Figure 4. Free-energy profile (ΔG, kcal/mol, b3-lyp/TZVPP//b3-lyp/TZVP) for the reaction between (1)CoPh and benzyl chloride (BzCl).
temporary spin flip even though there is no obvious need for one; this is unlike, for example, the halide abstractions described by Soper,2a where the ligand functions as an electron reservoir and the metal (CoIII) appears to be truly innocent. It is not clear at present how essential these details are to the different reaction outcomes. C−C Coupling Reactions. Since “standard” oxidative addition is so important in metal-catalyzed C−C coupling reactions, we decided to investigate whether the alkyl and aryl complexes studied here can also be used to effect C−C coupling reactions. Complexes (1)CoAr and (1)CoR were found to be much less reactive than (1)Co(N2) toward organic halides; only activated halides (allyl, benzyl, or iodides) reacted at all (Table 3).14 Complexes (1)CoAr bearing electron-withdrawing substituents at the Ar group are further deactivated, to the extent that, for example, complex 4 (formed from 2,6-dichloropyridine) is slower to react with benzyl chloride (BzCl) than even (1)CoCl.
Table 3. Reaction of Mixtures (1)CoR/(1)CoCl with Activated Alkyl Halides R′Xa,b products detected (%)c entry
R
R′X
RR′
RR
R′R′
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
p-CF3C6H4 C6H5 p-MeOC6H4 p-ClC6H4 p-CF3C6H4 C6H5 C6H5 π-crotyl CH2SiMe3e p-MeC6H4g C6H5 C6H5f p-CF3C6H4 p-MeOCOC6H4 6-Cl-2-C5H3N 6-Cl-2-C5H3N
CH2CHCH2Cl CH2CHCH2Cl CH2CHCH2Cl C6H5CH2Cl C6H5CH2Br C6H5CH2Br C6H5CH2Cl C6H5CH2Cl MeI MeI C6H5I n-C6H13Br n BuCl C6H5CH2Cl C6H5CH2Cl C6H5CH2Brg
77 84 94 50 23 55 77 100 100 93 n.r. n.r. n.r. n.o. n.o. trace
23 16 6 n.o. n.o. 39 22 n.o. n.o. 2σ GOF no. params R (Fo > 4σ(F)) R (all data) wR2 (all data) largest peak, hole/e Å−3
(1)CoNCH-4-C6H4Cl
(1)Co(η3-allyl)
(tBu2-2)CoI (Cl)
(tBu2-2)CoI2
C32H32ClCoN4 567.00 293(2) orthorhombic P212121 10.9905(8) 14.5602(11) 18.7056(15) 90 90 90 2993.3(4) 4 1.258 0.689 1184 −13 ≤ h ≤ 13 −17 ≤ k ≤ 17 −22 ≤ l ≤ 22 51 19 632 5568 3614 0.923 349 0.0523 0.0784 0.1443 0.421, −0.350
C28H32CoN3 469.50 293(2) orthorhombic Pccn 19.1003(11) 30.6877(18) 8.3627(5) 90 90 90 4901.7(5) 8 1.272 0.720 1984 −22 ≤ h ≤ 23 −37 ≤ k ≤ 37 −10 ≤ l ≤ 10 51 27 619 4558 2816 0.992 304 0.0435 0.0930 0.1444 0.363, −0.265
C37H53CoN3I0.94Cl0.06 720.17 293(2) monoclinic P21/n 13.586(5) 16.687(7) 16.630(6) 90 101.567(9) 90 3694(2) 4 1.295 1.282 1495 −16 ≤ h ≤ 16 −20 ≤ k ≤ 20 −20 ≤ l ≤ 20 51 26 916 6870 5955 1.159 395 0.0408 0.0494 0.1427 0.947, −0.765
C37H53CoI2N3 852.55 200(2) orthorhombic P212121 12.9841(8) 16.2444(10) 18.4949(11) 90.00 90.00 90.00 3900.9(4) 4 1.452 2.050 1716 −17 ≤ h ≤ 17 −20 ≤ k ≤ 21 −24 ≤ l ≤ 240 56.68 71 802 9576 8231 1.015 415 0.0344 0.0445 0.0979 1.472, −0.507
X-ray Structure Determinations. Crystal data and refinement parameters for the complexes are listed in Table 4. Details of individual structure determinations follow. (1)CoNCH-4-C6H4Cl. A deep purple fragment (approximately 0.20 × 0.20 × 0.25 mm) broken from a large crystal cluster was mounted in a thin glass capillary. Data were collected at 293 K in a Bruker four-circle diffractometer with an APEX detector using Mo Kα radiation (0.71073 Å). A sphere of data was collected with 0.2° scan width and 45 s scan time. The crystal system and space group were determined from the cell metrics and systematic absences. The data were integrated using the SAINT program,39 and a semiempirical absorption correction was done using SADABS.40 The structure was solved by Patterson methods using SHELXS41 and refined using SHELXL9741 (full-matrix least-squares refinement on F2); hydrogen atoms were placed at calculated positions and refined in riding mode. The structure was checked for solvent-accessible voids with PLATON.42 (1)Co(η3-allyl). A deep-orange crystal fragment (0.05 × 0.10 × 0.30 mm) was broken from a large piece of a crystalline aggregate and was sealed in a thin glass capillary and mounted on a Bruker D8 three-circle diffractometer equipped with a rotating anode generator (Mo Kα X-radiation), multilayer optics incident beam path, and an APEX-II CCD detector. A hemisphere of X-ray diffraction data (81 337 reflections) was collected to 60° 2θ using 25 s per 0.2° frame with a crystal-to-detector distance of 5 cm. A semiempirical absorption correction (SADABS) was applied, and identical data were merged to give 44 010 reflections covering the Ewald hemisphere, of which all data with up to 2θ = 51° were used for further refinement. The unitcell parameters were obtained by least-squares refinement of 4833 reflections with I > 10σ(I). (tBu2-1)CoI. For details of this (inconclusive) structure determination, see the SI. (tBu2-2)CoI (Cl). A deep-brown fragment (0.3 × 0.4 × 0.6 mm) was broken from a large piece of a crystalline aggregate and was sealed in a
(a) The mother liquor was evaporated to dryness, and the residue dissolved in toluene. The resulting solution was allowed to slowly evaporate in a drybox at room temperature. After four days, some dark cubes of a crystalline solid were obtained. A fragment broken from one of these crystals was used for singlecrystal X-ray diffraction, which showed it to be (tBu2-2)CoI contaminated with about 6% of (tBu2-2)CoCl. (b) The 1H NMR spectrum of the solid showed mainly diamagnetic (tBu2-2)CoI (at least 70%) but also one or more paramagnetic compounds. The solid was washed with 0.2 mL of ether/hexane. The remaining solid was further extracted with ether (∼0.5 mL)/toluene (6 drops), and the resulting solution was slowly evaporated to dryness over two weeks at room temperature to give a crystalline solid. A fragment broken off this solid was used for single-crystal X-ray diffraction, which showed it to be (tBu2-2)CoI2. For (tBu2-2)CoI: 1H NMR (benzene-d6, 300 MHz): δ 8.37 (1H, br d, Py CH), 7.96 (1H, t, J 7.2, Ar p), 7.60 (1H, d, J 7.5, Ar m), 7.41 (1H, d, J 7.1, Ar m), 6.77 (1H, t, J 7.3, Ar p), 6.65 (2H, t, J 7.0, Ar m), 4.63 (1H, m, NAr CH2CH3), 3.46 (1H, m, NAr CH2CH3), 3.16 (1H, m, NAr CH2CH3), 2.73 (2H, m, NAr CH2CH3), 2.53 (1H, m, NAr CH2CH3), 2.21 (2H, m, NAr CH2CH3), 1.71 (t, 3H, Jav 7.4, NAr CH2CH3), 1.31 (t, 3H, Jav 7.4, NAr CH2CH3), 0.99 (t, 3H, Jav 7.4, NAr CH2CH3), 0.96 (t, 3H, Jav 7.5, NAr CH2CH3), 0.36 (s, 9H, CMe3), 0.31 (s, 9H, CMe3), −2.32 (1H, br d, Py CHtBu), −2.80 (1H, s, Py CHtBu), −3.01 (s, 3H, NCMe). For (tBu2-2)CoI2 (incomplete and tentative assignments; all peaks are broad): 1H NMR (benzene-d6, 300 MHz): δ 109.1 (1H, Py H3), 15.4 (9H, CMe3), 12.2, 11.1, −16.1 (9H, CMe3). Since neither (tBu2-2)CoI nor (tBu2-2)CoI2 could be obtained pure, no elemental analysis was attempted. 3969
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
■
thin glass capillary and mounted on a Bruker D8 three-circle diffractometer equipped with a rotating anode generator (Mo Kα Xradiation), multilayer optics incident beam path, and an APEX-II CCD detector. A full sphere of X-ray diffraction data (132 039 reflections) was collected to 2θ = 60° using 2 s per 0.3° frame with a crystal-todetector distance of 5 cm. A semiempirical absorption correction (SADABS) was applied, and identical data were merged to give 42 999 reflections covering the Ewald hemisphere, of which all data up to 2θ = 51° were used for further refinement. The unit-cell parameters were obtained by least-squares refinement of 9940 reflections with I > 10σ(I). From the refinement results it became clear some (tBu2-2)CoCl had cocrystallized with the (tBu2-2)CoI (the Cl occupation refined to 6.3%). The source of this impurity is likely some (tBu2-2)CoCl present in the starting material (tBu2-2)CoCH2SiMe3. (tBu2-2)CoI2. A deep purple crystal block (0.16 × 0.19 × 0.26 mm) broken from a large piece of crystal was selected under an inert atmosphere and mounted on a glass fiber. Unit cell measurements and intensity data collections were performed on a Bruker-AXS SMART 1K CCD diffractometer using graphite-monochromatized Mo Kα radiation (λ = 0.71073 Å). The measurement was done at 200 K. The data reduction included a correction for Lorentz and polarization effects, with an applied multiscan absorption correction (SADABS). The crystal structures were solved and refined using the SHELXTL program suite. Direct methods yielded all non-hydrogen atoms, which were refined with anisotropic thermal parameters. All hydrogen atom positions were calculated geometrically and were riding on their respective atoms. Thermal parameters of carbon atoms C26, C27, and C37 suggested the presence of positional disorder. Disorder for C26 and C27 was modeled as oscillation of the ethyl substituent moiety with partial occupancies of 50%:50%. Disorder for C37 was modeled as oscillation of the methyl end group of another ethyl substituent with partial occupancies of 50%:50%. A set of “rigid-bond” and “thermal parameters” restraints was applied to the disordered fragments to improve refinement of thermal parameters. Computational Details. All geometries were optimized with Turbomole43 using the TZVP basis set,44 the b3-lyp functional,45 and the unrestricted DFT formalism in combination with an external optimizer (PQS OPTIMIZE).46 The low-spin state was found to be the lowest in energy for most species studied; square-planar Co(I) complexes preferred a broken-symmetry Sz = 0 solution.20 Vibrational analyses were done to confirm the nature of all stationary points and to calculate thermal corrections (enthalpy and entropy, gas phase, 298 K, 1 bar) using the standard formulas of statistical thermodynamics. Improved single-point energies were obtained using the TZVPP basis set47 at TZVP geometries and combined with TZVP-level thermal corrections to generate the final free energies.
■
REFERENCES
(1) (a) Johnson, L. K.; Killian, C. M.; Brookhart, M. J. Am. Chem. Soc. 1995, 117, 6414−6415. (b) Wissing, E.; Kaupp, M.; Boersma, J.; Spek, A. L.; Vankoten, G. Organometallics 1994, 13, 2349−2356. (c) Fedushkin, I. L.; Tishkina, A. N.; Fukin, G. K.; Hummert, M.; Schumann, H. Eur. J. Inorg. Chem. 2008, 483−489. (d) Scarborough, C. C.; Wieghardt, K. Inorg. Chem. 2011, 50, 9773−9793. (2) (a) Smith, A. L.; Clapp, L. A.; Hardcastle, K. I.; Soper, J. D. Polyhedron 2010, 29, 164−169. (b) Verma, P.; Weir, J.; Mirica, L.; Stack, T. D. P. Inorg. Chem. 2011, 50, 9816−9825. (3) (a) Gibson, V. C.; Redshaw, C.; Solan, G. A. Chem. Rev. 2007, 107, 1745−1776. (b) Gibson, V. C.; Solan, G. A. Iron-based and Cobalt-based Polymerisation Catalysts. In Metal Catalysts in Olefin Polymerization; Guan, Z., Ed.; Springer: Berlin, 2009; Vol. 26, pp 107− 158. (4) Chirik, P. J.; Wieghardt, K. Science 2010, 327, 794−795. (5) Bart, S. C.; Lobkovsky, E. B.; Chirik, P. J. Abstr. Pap. Am. Chem. Soc. 2004, 228, U898−U898. (6) Spessard, G. O.; Miessler, G. L. Organometallic Chemistry, 2nd ed.; Oxford University Press: Oxford, 2010. (7) (a) Trovitch, R. J.; Lobkovsky, E.; Chirik, P. J. J. Am. Chem. Soc. 2008, 130, 11631−11640. (b) Trovitch, R. J.; Lobkovsky, E.; Bouwkamp, M. W.; Chirik, P. J. Organometallics 2008, 27, 6264−6278. (8) (a) Doherty, J. C.; Ballem, K. H. D.; Patrick, B. O.; Smith, K. M. Organometallics 2004, 23, 1487−1489. (b) MacLeod, K. C.; Conway, J. L.; Tang, L. M.; Smith, J. J.; Corcoran, L. D.; Ballem, K. H. D.; Patrick, B. O.; Smith, K. M. Organometallics 2009, 28, 6798−6806. (9) (a) Halpern, J.; Maher, J. P. J. Am. Chem. Soc. 1965, 87, 5361− 5366. (b) Chock, P. B.; Halpern, J. J. Am. Chem. Soc. 1969, 91, 582−&. (c) Schneider, P. W.; Phelan, P. F.; Halpern, J. J. Am. Chem. Soc. 1969, 91, 77−. (d) Marzilli, L. G.; Marzilli, P. A.; Halpern, J. J. Am. Chem. Soc. 1971, 93, 1374−. (e) Halpern, J.; Phelan, P. F. J. Am. Chem. Soc. 1972, 94, 1881−1886. (10) Ogoshi, H.; Setsune, J.; Yoshida, Z. J. Am. Chem. Soc. 1977, 99, 3869−3870. (11) Zhu, D.; Budzelaar, P. H. M. Organometallics 2010, 29, 5759− 5761. (12) Smith, A. L.; Hardcastle, K. I.; Soper, J. D. J. Am. Chem. Soc. 2010, 132, 14358−14360. (13) Enright, D.; Gambarotta, S.; Yap, G. P. A.; Budzelaar, P. H. M. Angew. Chem., Int. Ed. 2002, 41, 3873−3876. (14) Zhu, D.; Thapa, I.; Korobkov, I.; Gambarotta, S.; Budzelaar, P. H. Inorg. Chem. 2011, 50, 9879−9887. (15) (a) Cheung, C. W.; Chan, K. S. Organometallics 2011, 30, 4999−5009. (b) Cheung, C. W.; Chan, K. S. Organometallics 2011, 30, 4269−4283. (c) Cheung, C. W.; Chan, K. S. Organometallics 2011, 30, 1768−1771. (16) Bowman, A. C.; Milsmann, C.; Atienza, C. C. H.; Lobkovsky, E.; Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc. 2010, 132, 1676−1684. (17) (a) Knijnenburg, Q.; Horton, A. D.; van der Heijden, H.; Kooistra, T. M.; Hetterscheid, D. G. H.; Smits, J. M. M.; de Bruin, B.; Budzelaar, P. H. M.; Gal, A. W. J. Mol. Catal. A: Chem. 2005, 232, 151−159. (b) Tellmann, K. F.; Humphries, M. J.; Rzepa, H. S.; Gibson, V. C. Organometallics 2004, 23, 5503−5513. (18) PhCHCHPh and PhCH2CH2Ph detected by GC/MS after hydrolysis. (19) (a) Erker, G.; Fromberg, W.; Atwood, J. L.; Hunter, W. E. Angew. Chem., Int. Ed. Engl. 1984, 23, 68−69. (b) Kaplan, A. W.; Polse, J. L.; Ball, G. E.; Andersen, R. A.; Bergman, R. G. J. Am. Chem. Soc. 1998, 120, 11649−11662. (c) Uhl, W.; Matar, M. Z. Naturforsch. Sect. B: J. Chem. Sci. 2004, 59, 1214−1222. (d) Temprado, M.; McDonough, J. E.; Mendiratta, A.; Tsai, Y. C.; Fortman, G. C.; Cummins, C. C.; Rybak-Akimova, E. V.; Hoff, C. D. Inorg. Chem. 2008, 47, 9380−9389. (20) Knijnenburg, Q.; Hetterscheid, D.; Kooistra, T. M.; Budzelaar, P. H. M. Eur. J. Inorg. Chem. 2004, 1204−1211. (21) The (3)Co(N2) used for this reaction was prepared via Na/Hg reduction of (3)CoCl2.
ASSOCIATED CONTENT
S Supporting Information *
CIF file containing X-ray structure data for the structures determined in this work; total energies and xyz coordinate files for all optimized geometries. This material is available free of charge via the Internet at http://pubs.acs.org.
■
Article
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Mr. Mark Cooper (University of Manitoba) for help with X-ray structure determinations and Prof. Frank Hawthorne for the use of their single-crystal X-ray diffractometer. This work was supported by NSERC, CFI, and MRIF grants (to P.H.M.B.) and a University of Manitoba Graduate Fellowship and Manitoba Graduate Scholarship (to D.Z.). 3970
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971
Organometallics
Article
(22) Note that since LCoAr and LCoX are formed in a ratio
